File viewer renders *.c file as an image with Pygments 2.11 - 2.12

[Jun Omae]

Reported at ​​gmessage:trac-users:iOcdoLFjKZg/m/IsjjrUZgEgAJ.

[Jun Omae]

Root cause is ​https://github.com/pygments/pygments/commit/1f44b277737f88a4402032584660d84a671fb79d

[Jun Omae]

PygmentsRenderer.get_extra_mimetypes() with Pygments 2.11+ returns multiple mimetypes for c.

>>> from trac.test import EnvironmentStub
>>> from trac.mimeview.api import Mimeview
>>> from trac.mimeview.pygments import PygmentsRenderer
>>> env = EnvironmentStub()
>>> mimeview = Mimeview(env)
>>> mimeview.mime_map['c']
'image/x-xpixmap'
>>> mimeview.get_mimetype('filename.c')
'image/x-xpixmap; charset=utf-8'
>>> renderer = PygmentsRenderer(env)
>>> list(renderer.get_system_info())
[('Pygments', '2.12.0')]
>>> for mimetype, kwds in renderer.get_extra_mimetypes():
...   if 'c' in kwds: print(mimetype, kwds)
...
text/x-chdr ('c',)
text/x-csrc ('c',)
image/x-xbitmap ('c',)
image/x-xpixmap ('c',)

When Pygments 2.10.0 is used:

>>> mimeview = Mimeview(env)
>>> renderer = PygmentsRenderer(env)
>>> list(renderer.get_system_info())
[('Pygments', '2.10.0')]
>>> mimeview.mime_map['c']
'text/x-csrc'
>>> mimeview.get_mimetype('filename.c')
'text/x-csrc; charset=utf-8'
>>> for mimetype, kwds in renderer.get_extra_mimetypes():
...   if 'c' in kwds: print(mimetype, kwds)
...
text/x-chdr ('c',)
text/x-csrc ('c',)

Work around is to avoid one alias having multiple mimetypes like this:

trac/mimeview/pygments.py

@@ -109 +109 @@
         for _, aliases, _, mimetypes in get_all_lexers():
             for mimetype in mimetypes:
                 yield mimetype, aliases
+                aliases = ()  # avoid one alias having multiple mimetypes
 
     def get_quality_ratio(self, mimetype):
         # Extend default MIME type to mode mappings with configured ones

After the changes:

>>> from trac.test import EnvironmentStub
>>> from trac.mimeview.api import Mimeview
>>> from trac.mimeview.pygments import PygmentsRenderer
>>> env = EnvironmentStub()
>>> mimeview = Mimeview(env)
>>> mimeview.mime_map['c']
'text/x-chdr'
>>> mimeview.get_mimetype('filename.c')
'text/x-chdr; charset=utf-8'
>>> for mimetype, kwds in renderer.get_extra_mimetypes():
...   if 'c' in kwds: print(mimetype, kwds)
...
text/x-chdr ('c',)

[Jun Omae]

Proposed changes in [a5d3519b1/jomae.git] (jomae.git@t13483). After the changes, text/* type will be preferred in order to prevent rendering as an image for *.c filename.

[Jun Omae]

Pushed in [17623].